This release adds 3 notable features for engineering teams evaluating rollout.
✓ No known CVEs patched in this version
Topics
+7 more
Affected surfaces
Summary
AI summaryNew hosted mode enables stateful orchestration without API keys, adds debate style options, context injection, and structured synthesis output.
Full changelog
What's New
Hosted Mode (No API Keys!)
New mode="hosted" where the MCP server acts as a stateful prompt orchestrator — it generates debate prompts but never calls model APIs. The host (Claude Code, Copilot, etc.) executes prompts using its own model access via sub-agents. Works with opus/sonnet/haiku in Claude Code. Same model can be used multiple times for diverse perspectives.
New brainstorm_collect tool for the host to submit model responses back to the server.
Debate Styles
New style parameter:
- freeform (default) — open brainstorming
- redteam — adversarial fault-finding, models find flaws, risks, and weaknesses
- socratic — probing questions that expose assumptions and push for deeper understanding
Context Injection
New context parameter to ground debates in actual code snippets, PR diffs, error logs, or architecture docs. Models debate YOUR work, not abstract questions.
3-Bullet Synthesis Verdict
Upgraded synthesis output to structured format:
- Recommendation — one clear, opinionated recommendation
- Key Tradeoffs — 2-3 most important considerations
- Strongest Disagreement — the sharpest unresolved tension (not watered down)
VS Code Copilot Support
.agent.mdfiles for model-specific brainstorm agents.vscode/mcp.jsonconfiguration for Copilot MCP integration- Tool asks user to choose API vs Hosted mode when not specified
Mode Selection UX
When mode is not provided, the tool returns a prompt asking the user to choose between API and Hosted mode — ensures the right flow every time.
Weekly OSS security release digest.
The CVE patches and breaking changes that affected production tools this week. One email, every Sunday.
No spam, unsubscribe anytime.
Share this release
About spranab/brainstorm-mcp
Multi-round AI brainstorming debates between multiple models (GPT, Gemini, DeepSeek, Groq, Ollama, etc.). Pit different LLMs against each other to explore ideas from diverse perspectives.
Related context
Beta — feedback welcome: [email protected]